Blackwell supercomputer for medical AI in the Netherlands
One of the first Blackwell supercomputers in the Netherlands has been fired up at a datacentre in Rotterdam for medical AI applications.
The Nvidia DGX B200 system at the NorthC datacentre is being used by startup Juvoly for an AI that understands the Dutch language for voice-enabled speech-to-text healthcare workflows. Blackwell is the latest GPU technology from Nvidia, overtaking the existing Hopper architecture in shipments.
Juvoly is collaborating with Dutch universities Erasmus MC and TU Delft to develop a Medical Dutch Large Language Model (LLM), which is planned to be released as open-source in the future.
- Nvidia preps Nvidia Ultra, next generation chips
- Blackwell pitched as the start of a while new industry
- Nvidia doubles in AI boom
The B200 (above) has eight Nvidia Blackwell GPUs interconnected and two Intel Xeon CPUs with fifth-generation NVLink interconnect to provide 70 petaFLOPS of AI training performance, three times the previous generation, and 11petaFLOPS of inference, up 15x. The system was unveiled by Prince Constantijn van Oranje who is a special envoy for the Dutch deep tech startup ecosystem.
“With this infrastructure, we can develop AI solutions that improve healthcare outcomes, without sensitive patient data and other information being sent abroad,” says Thomas Kluiters, co-founder of Juvoly. “If a startup less than two years old can bring the latest supercomputers to the Netherlands, then other companies can too.”
The company creates and hosts speech-to-text models optimized for language barriers, accents, and medical terminology with a model that outperforms international players while consuming less energy. It is already processing over 5,000 general practitioner consultations per day.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
